Change detection

Change detection

The Change Detection is a technique for processing images obtained using satellite remote sensing technology, and in our case at high resolution, for identifying and/or discriminating changes in the environment over time. In updating the analyses of the territory, the E.R.O.S. A Mission data, acquired by the Multimission Satellite Station located at Villagrande Strisaili (NU) in Sardinia, are georeferenced and orthorectified on the basis of the same previously used aerial photo reference system .

The Change Detection represents a major breakthrough in keeping track of variations that otherwise would not be perceivable from official documentation.

The EROS A satellite images used offer true and updated information thanks to the high degree of geometric detail (GSD 1.8mm). The use of such products has enabled us to monitor the territory, whether urban or not, at regular time intervals. This enables us to follow developments in the territory over time. In studying temporal changes, the following methodologies have been used:

Display in RGB
The individual images, previously orthorectified, are allocated to each individual display channel corresponding to the RGB system. The colour combination displayed indicates variations, if any, that have taken place over time. The areas that have not undergone any change remain a shade of grey, the shade varying according to the intensity of the colour of each individual area. However, if there have been changes in the territory, such changes will be displayed in hues of colour that vary according to the radiometric intensity of the area and to the type of change that has taken place over time. Variations in colour may also provide a qualitative assessment of the type of change that has occurred .
Contemporaneous classification
This is concerned with grouping together the thematic characters of the image and defining the degree of pertinence of each element of the image to prearranged classes. In addition to radiometry, information concerning space and structure are also introduced into this classification system. Once the supervised classification is completed, it is then possible to make a comparison via matrixes that result in a valuation of the degree of variation between the images.

Segmentation
This consists of carrying out a classification, in homogeneous areas, of images and analyzing the results obtained via the simultaneous display or production of an image of the differences. The images are then transformed into vectorial format and also made available in the GIS environment used. The vectorialization of the images ensures complete compatibility with other data to be inserted for a complete assessment of the impact on the environment.

The difference between the methods described above have permitted a progressive reduction in the need for an operator with the gradual automation of processes, and also a progressive improvement in the qualitative and quantitative accuracy of the results. The uniqueness of the solution, proposed by IPT, is demonstrated by the incorporation of paper and digital cartographic and cadastral map information, with digital images of the territory received directly from a satellite via its receiving station in Italy.